Font Size: a A A

Rough Decision Rules Reduction Based On Granular Computing

Posted on:2018-02-18Degree:MasterType:Thesis
Country:ChinaCandidate:F ZhangFull Text:PDF
GTID:2348330533470701Subject:Mathematics
Abstract/Summary:PDF Full Text Request
In general,based on rough set theory,the rule mining is a process where the decision table undergoes attribute reduction and then the decision rules are attained through the reducing attribute values.For the decision information system with multi valued attribute,the problem was solved by the kind of methods mostly through the uniformization of the multi value,increasing the complexity of the computation.Granular computing theory was applied in the reduction of multi valued attribute and puts forward a method of rule extraction and reduction based on rough set granular computing model,avoiding scanning decision tables frequently.The main research fruits are as follows.Firstly,granular membership function is presented based on the rough membership function and the distance between object attributes and center point.Each object in the decision table is mapped to the granular space,so the granulation of decision table is realized.The structure of granular layer and the properties of concept granular are studied.Secondly,the ternary structure of concept granule is defined based on the concept lattice theory of formal concept analysis.The construction method of the concept granular set is given through analyzing the granularity and decomposition and synthesis algorithm of the concept granule.Meanwhile,attribute resolution operator is defined and the calculation basis of attribute importance is given.Thirdly,the construction algorithm of concept granular lattice is given based on hierarchical constructing strategy of concept lattice.Attribute reduction is realized through isomorphism theory of lattice structure.After the reduction,the rules are extracted from the concept granular lattice.Finally,an empirical study was carried out on proposed method using the data of HSV cases in the UCI database as the application background,present of the method proposed by itself.They illustrate that computing model based on rough set granular,which has a good application value and bright prospect,is effective and practical.
Keywords/Search Tags:granular computing, multi valued attributes reduction, decision information system, concept granular lattice, rule reduction
PDF Full Text Request
Related items